John had a thought. [“a” to match “thought”]
John had an idea. [“an” to match “idea”]
Obviously the above are correct - but what about the following?
John had _ (very interesting) idea. [“a” to match “very”, or “an” to match “idea”?]
John had _ (in hindsight, very interesting) thought. [“an” to match “in”, or “a” to match “thought”?]
In both cases it should be “a” to match “very interesting”
It depends on the initial sound of the immediately following word. So it would be “a (very . . .” and “an (in . . .” in your examples.
Imagine yourself saying the sentence out loud. The parenthesis might be indicated by stress or timing or whatever, but whether you use “a” or “an” will be dictated by the whether the next succeeding word starts with a vowel, and that doesn’t depend on whether there is a parenthesis in between.
Thirded.
The second example is problematic, because in hindsight is not really part of the noun phrase. In hindsight is not an intrinsic attribute of the thought itself, but rather the speaker’s perspective. (A thought can’t look at itself in hindsight.)
“Normally” one does not insert a clause-level adverbial (in hindsight) between an article and the rest of a noun phrase. The normal English syntax–for planned speech–would have the speaker say, John had, in hindsight, a very interesting thought. If this is clearly intentional deviation from normal syntax–in order to call attention to some quality of the thought itself–then I could see the speaker using an, but I see it as clearly marked language.
However, the second sentence, as a representation of real-time spoken discourse, could also be a case of something that is called repair in discourse analysis–kind of like mentally backtracking, and altering one’s speech while one is speaking, with regard to something one has just said. In this case, perhaps it didn’t occur to the speaker to say “in hindsight” until after he or she had started the noun phrase. As such, there is no reason for the speaker to say an, because the following modifier (in hindsight) is not actually (indeed, cannot be) part of the noun phrase, and wasn’t initially planned.
Either way, it’s not actually about the subsequent sound–as some have said here–but rather the subsequent sound that is part of the actual noun phrase.
In most cases, the way I determine the use of ‘a’ or ‘an’ is this:
If the word following ‘a’ or ‘an’ in the sentence is a vowel, then I usually use ‘an’.
If the word following ‘a’ or ‘an’ in a sentence is a consonant, then I usually use ‘a’.
There are exceptions to that rule. It wouldn’t be English if rules weren’t broken.
Saying it in my head both ways helps a lot. I go with whichever sounds right.
Well, which sounds “more right” to you?
A) They live in a–I think–nice house.
B) They live in an–I think–nice house.
Because this pretty much is the same kind of thing in the OP. The noun phrase (a nice house) is getting “interrupted” by something that is syntactically extraneous.
I think the problem her is the structure of the sentence. It should be “I think they live in a nice house” or “I think the house they live in is nice” (or, if he really wants to put “they” first, “they live in a house which I think is nice”).
As you said yourself, what we have here is a “repair”, where the speaker realises he has started his sentence wrongly and interrupts himself to interject something he should have said earlier.
You wouldn’t normally replicate this in writing, except where you’re attributing this speech to someone and you want to convey that person’s error or confusion.
Given that the sentence construction it itself erroneous, I’m not sure if there’s much point in having a “rule” about which form of the indefinite article to use in a sentence which itself breaches the “rules”. A real-life speaker will use whichever form pops into his head as he realises his error and grapples with correcting it. Since he’s by definition confused and distracted from the issue of which form of the article to choose, we probably should hesitate to base any norms on the choice he makes.
The sample sentences are awkward because they are breaking the rule* that you should not separate the article from its’ noun.
So the ‘in hindsight’ and ‘I think’ should not be put in between, but moved before the article. Or the whole sentence should be reworded, which might be the better option.
But it’s perfectly normal (in fact, very productive) in English to place modifiers–adjectives and adverbs–between an article and its noun. (Where else are you going to put them?) In other words:[ul]
[li]That is a rather dubious rule.[/li][/ul]The issue here is, as UDS and I mentioned above, what kind of language you use to separate the article and its noun.
John had a ((in hindsight) very interesting) thought.
That’s better.
guizot, I agree with you that in normal speech the article would agree with the intended following word, even if another word is interpolated.
In writing, that’s not possible. The next actual word is immediately apparent. Additionally, some readers sound out the words mentally. Both require that the article agree with the adjacent word sound or else it feels wrong.
That may be the intention. If the writer were intending to accurately capture the flow of speech, articles as well as tenses and plurals might be off because the sentence did not finish as originally intended. That’s a big if. I’d say that 99% of speech in fiction adheres to an idealized and grammatically correct version of speech if an idiolect is not being portrayed.
We were not told the context of the lines by the OP. As presented, as a written sentence, the article must agree with the following sound. That makes it different from your examples in which the dashes create a mental break that does allow for the inserted words to be separated from the word flow. Your examples are correct in their very different context. They would not be correct in the OP’s context.
I think the parenthesis are wrong.
“John had a very interesting idea.”
“John had, in hindsight, a very interesting thought.”
I don’t think a preposition takes an article in English.
Sure–as I said, in the case of a speaker willfully employing in hindsight as a modifier of very interesting, which it grammatically is not, since it is a clausal modifier—that is, if the speaker is “repurposing“ the phrase deliberately–then it makes sense to say an. Whether or not one considers that “idealized and grammatically correct” is a question of personal preference, I suppose.
I searched COCA and couldn’t find any examples of in hindsight being used after an article, or as the strict modifier of an adjective or adverb of a noun phrase. It’s used almost exclusively as a clausal modifier in cases such as this:*When the war ended, we supported – in hindsight too unquestioningly – a cease-fire agreement that left Saddam Hussein in power.*I think we can see that it is such a common and convenient clausal phrase, though, that it has clearly taken on the kind of adverbial function that would lend itself to the kind of noun phrase in the OP’s example. By way of prosody the speaker can easily “repurpose” it in this way.
No, nested parentheses are never better, except maybe for mathematicians and computer scientists.
Now I’m having flashbacks to college, when I had a math/CS professor who loved the programming language Scheme.
To the OP’s question, I too agree with answers like UDS’s “Imagine yourself saying the sentence out loud.” And if you can’t imagine yourself saying it out loud, in a not totally awkward way, you might want to rethink the sentence.
The entire purpose of adding an “n” to the indefinite article “a” is to facilitate speech. When spoken, “an apple” sounds much less awkward than “a apple.” And in speech, there are no parentheses; the words just flow one after the other. A parenthesis may change the inflection of the sentence, but the sounds proceed uninterrupted. The grammar of the sentence is totally irrelevant. You’re still following “a” with a vowel sound, which would change it to “an.” So, in keeping with this, “an (in hindsight . . . )” is correct.
Use of an vs. a is not a grammatical issue, since the words are grammatically identical. It doesn’t matter which word the article is referring to. It is entirely a pronunciation issue. Which one to use depends on how the following word is pronounced. If it begins with a vowel sound, use an, if not, use a. That’s really the only rule that applies here.
In general I agree, but keep in mind that the reason “a apple” sounds so awkward is because you never hear it in that context. That is, the rule about this kind of vowel adjacency is specific to the indefinite article. In other contexts, we don’t have a problem with that particular vowel adjacency in the OP–for example, with:Hosanna in the highest(well, at least if you’re a good Catholic).
In other words, there is no larger phonological rule which makes that vowel adjacency prohibited throughout English–it just applies here, with the indefinite article. For this reason, I would argue that a speaker may very well override the phonological rule in this particular context in order to convey the specific semantic implication of the grammatical exception. In real-life speech people do this all the time. (Whether to represent this in print is a different matter.)